[tbb-bugs] #20352 [Applications/Tor Browser]: Integrate sandboxed Tor Browser into our gitian build system
Tor Bug Tracker & Wiki
blackhole at torproject.org
Thu Nov 24 12:36:40 UTC 2016
#20352: Integrate sandboxed Tor Browser into our gitian build system
-------------------------------------------------+-------------------------
Reporter: gk | Owner: tbb-
| team
Type: task | Status:
| needs_review
Priority: High | Milestone:
Component: Applications/Tor Browser | Version:
Severity: Normal | Resolution:
Keywords: tbb-gitian, tbb-sandboxing, | Actual Points:
GeorgKoppen201611, TorBrowserTeam201611R |
Parent ID: #19750 | Points:
Reviewer: | Sponsor:
-------------------------------------------------+-------------------------
Comment (by gk):
Replying to [comment:16 boklm]:
> Replying to [comment:11 gk]:
> > I have bug_20352_v4 (https://gitweb.torproject.org/user/gk/tor-
browser-bundle.git/commit/?h=bug_20352_v4 commits
03852126e8abb782c850dc9230eaf98e81851366 and
b7f4ab23fe3e508fee6eab106632679992ca4688) which is up for review (while I
look at the reproducibility issues). boklm: Could you please have a look
and test it on your machine(s)? One thing I am particularly interested in
is your report about whether killing the old qemu process (from the
previous build step) still works as it used to. It turns out on my machine
I have to wait about an hour [sic!] until the process gets killed. If that
is not just my machine I'd be happy to hear about possible workarounds or
fixes.
>
> I tried building that on my machine, but it gets stuck while creating
the VMs:
> {{{
> Formatting 'target-jessie-amd64.qcow2', fmt=qcow2 size=17179869184
backing_file='base-jessie-amd64.qcow2' encryption=off cluster_size=65536
lazy_refcounts=off
> }}}
>
> I can see this qemu process running:
> {{{
> qemu-system-i386 -enable-kvm -m 2000 -smp 2 -drive file=target-
jessie-i386.qcow2,cache=writeback,if=virtio -net nic,model=virtio -net
user,hostfwd=tcp:127.0.0.1:2223-:22 -vnc 127.0.0.1:16
> }}}
>
> After stopping the build, killing qemu, and restarting it, it gets stuck
again at the same point.
That's weird because after the first try you should have a `base-
jessie-i386.qcow2` and during the second one (after killing qemu) you
would create the 64bit one. Running `./make-vms.sh` again now should
indicate that all images are created. Maybe you are seeing something
different like the build is going trying to kill the qemu process dealing
with jessie 64bit?
Either way it seems you hit the same problem as I: the images are not
shutting down properly. Any idea how to cope with that? Maybe we kill them
right away in `stop-target` until we have a proper fix at least?
--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/20352#comment:17>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online
More information about the tbb-bugs
mailing list